II / CRYSTALLIZATION / Additives: Molecular Design
CRYSTALLIZATION
Additives: Molecular Design
J. H. ter Horst and R. M. van Rosmalen,
Delft University of Technology, The Netherlands
R. M. Geertman, Akzo Nobel Chemicals Research,
Arnhem, The Netherlands
Copyright ^ 2000 Academic Press
Introduction
Crystallization is the nucleation and growth of a solid
phase in a mother liquor. It can be used as a separ-
ation technique, because the process results in the
separation of the mother liquor and the solid phase. If
the mother liquor and the solid phase have differ-
ent compositions it can also be used as a puriRcation
technique. In such cases the mother liquor must con-
tain impurities. The deliberate addition of impurities
during the crystallization process, has been widely
used to improve both the product quality and the
process performance. However, most impurities pres-
ent during crystallization processes are not added on
purpose, but are by-products from previous reac-
tions, dissolved salts and solvents that can have the
same kinds of effects.
In industrial crystallization additives are generally
used to improve the handling characteristics of the
crystalline product or to prevent scaling. By employ-
ing correctly selected additives it is possible to reduce
caking, improve the free Sowing behaviour of a
powder, increase the bulk density of a crystalline
product, change the crystal size distribution,
induce certain polymorphs and improve Rltration
behaviour.
Varied though this list may seem, all these ef-
fects can be achieved by using additives that have very
speciRc interactions with the crystal surface. This
fundamental physical effect causes a change in
the system parameters when additives are introduced.
Detailed knowledge of the molecular structure of
the solid}liquid interface is very important for design
of additives. Once this interface is understood, it
is possible to design molecules that have speciRc
interactions with one or more of the faces of a crystal.
Often the same additive can be used in different
application. For instance a growth retarder used to
prevent scaling of a given compound can, when ad-
ded after crystallization, also act as an anticaking
931
agent. Blocking a single crystal face with a certain
additive can induce morphology changes, but the
same additive can also be used to act as a template for
the nucleation of that same crystal face.
In many cases additives are not wanted in the
product application. Ferrocyanide is a very effec-
tive anticaking agent for salt (NaCl), but the ferro-
cyanide complex must be destroyed and the iron
precipitated prior to electrolysis of the salt for the
production of chlorine (Cl2). Nitrosyltriacetamide is
also an effective anticaking agent for salt, but
during electrolysis nitrogen trichloride (an explosive)
is formed.
Apart from having a very speciRc inSuence on
a crystal face, additives must also be effective at
very low concentrations. Because one of the main
uses of additives is in cheap bulk products (costing
less than US $1 kg\1) the price of the additive also
acts as an incentive to keep the additive concentra-
tions as low as possible. Expensive additives, even if
used in very low amounts, will adversely affect
the cost of these cheap bulk products. High additive
concentrations always give rise to signiRcant additive
incorporation in the crystal and thus decrease the
product purity.
Additives that effectively control crystallization
processes are also found in nature: the shells of crus-
taceans are formed as a result of bio-templated
growth of calcium carbonate; the growth of the min-
eral components of our bones and teeth is carefully
regulated; and studies of Rsh in polar regions have
revealed some proteins that very effectively
block the growth of ice crystals.
After describing the fundamental effects of the
additive in the crystallization system, a general
procedure for the molecular design of additives
will be given. The review ends with some case
studies.
Interaction of Impurities and
Crystal Surfaces
The effect of an impurity that is present in small
quantities (parts per million (ppm) level) cannot be
explained solely by its effect on solution bulk
parameters such as solubility. The phenomenon
that is at the root of all the effects of impurities
on process performance and product quality is
the interaction of the impurity with the crystal
surfaces.
932 II / CRYSTALLIZATION / Additives: Molecular Design
Adsorption of the impurity can take place on Sat
surfaces, at stepped surfaces and at kink sites (Fig-
ure 1). The adsorbed compound can impede the
movement of steps on the surface. This is a kinetic
effect and causes a reduction in the growth rate
of the corresponding crystal surface. For instance the
induction time, i.e. the time that elapses until the
outgrown nuclei are detectable, can increase because
the outgrowth of nuclei takes longer.
Origin and Type of Impurities
In solution crystallization the solvent is an inevitable
impurity. Additives play a role in the improvement
of product quality (e.g. shape of the crystals,
crystal size distibution) or process performance (e.g.
Rlterability). One source of impurities in the crystalli-
zation process is the synthesis of the crystal com-
pound, as synthesis by-products greatly affect
crystallization.
Impurities can also be classiRed based on their
character. Apart from organic and inorganic com-
pounds, a special group of impurities is the polyelec-
trolytes, long carbon chain molecules with many
functional groups. Because of the large number of
functional groups the interaction between polyelec-
trolytes and crystal surfaces is strong. The large num-
ber of functional groups also makes the molecule
kinetically very hard to remove, as all the bonded
functional groups have to be detached from the sur-
face. One of the detached functional groups remain-
ing bonded to the polyelectrolyte and is thus posi-
tioned very near the crystal surface. The attachment
of that one functional group onto the surface is faster
than the detachment from the surface of the other
functional groups of the polyelectrolyte molecule.
The strong interaction and the large number of inter-
action points make polyelectrolytes highly effec-
tive additives at very low concentrations. Another
special group of impurities is the crystallizing com-
pounds themselves. Long chain hydrocarbons such as
fats may cause self-poisoning on certain surfaces if
the fat molecules in the melt near the surface have
a different conformation from the fat molecules
in the surface.
Figure 1 Additive adsorption at various positions on a crystal
surface: 1, flat surface; 2, step; 3, kink site.
Predicting the Effects of
Impurities
The deliberate addition of an impurity is intended to
cause a particular effect, such as a reduction in
mineral scale formation or a change in morphology.
This effect comes from changes in the kinetic
process involved, e.g. for a change in morphology the
relative growth rates of the different crystal faces
have to change. The function of the additive is thus
very speciRc: the morphology changes if the growth
of a speciRc face is blocked. This is illustrated in
Table 1 for a number of desired additive effects.
The Rrst column gives the desired effect, the
second the kinetic process involved and the third the
action acquired by the additive. Table 2 shows
whether an interaction with a speciRc face or all faces
is required, at what particular moment the additive
should be added and whether it blocks a face or acts
as a template. This table shows the close relationship
between additives used for different purposes.
An anticaking agent is simply an antiscaling agent
added after crystallization, and an additive used to
block a speciRc crystal face can in some cases also act
as a template for that same face.
IdentiRcation of an unintentional impurity and un-
derstanding of its effect on the crystallization
process requires a kind of reverse engineering. Many
impurities are formed as by-products during the syn-
thesis of the compound to be crystallized. By observ-
ing how the crystallization process changes in the
presence of impurities, the kinetic process involved
can be identiRed. The effect on the process then
can be matched with the expected (predicted) ef-
fect of one of the impurities present. This is a rather
complicated procedure because not all the impurities
present that can act on the kinetic processes involved
will be known.
The desired effects of additives are changes in
the crystal quality or the process parameters. The
crystal quality is an overall term for a variety of
crystal parameters (morphology, purity, etc.).
Changes in the process conditions due to additives,
such as better Rlterability, make the process more
efRcient. Figure 2 shows the effects of the
impurities on the crystal quality and the process con-
ditions. The working directions for both the rational
design and the reverse engineering are shown. Key to
this Rgure is knowledge of the interaction of a given
compound with the crystalline interface. This interac-
tion determines whether an impurity can, e.g. block
growth or induce nucleation of a given face. After
evaluating the inSuence of a given impurity on the
morphologically most important faces, the overall
effect on the crystal growth and the outgrowth of
 II / CRYSTALLIZATION / Additives: Molecular Design 933

Table 1 Additive effects on the crystal growth process

Desired effect Kinetic process involved Action required

Anticaking Dissolution and regrowth of crystalline Complete blocking of growth

Antiscaling
Changing morphology (filtration, flowability)
Control of polymorphism
Changing crystal size distribution
material
Complete blocking of nucleation and/or
growth
Ratio of growth rates of different faces
Nucleation of a given polymorph
Changing nucleation/growth ratio
Complete blocking of growth
Block specific faces
Mimicing a face of the desired polymorph
Block specific faces, or block all faces

nuclei can be deduced. From these effects the
inSuence on the crystal quality and the product hand-
ling characteristics and ultimately the product prop-
erties can be determined. The starting point for de-
signing and studying additives is therefore a study of
the molecular structure of the crystalline interface.
Molecular Structure of the Crystal
Surface
Both the impurity and the crystal surface determine the
strength of the interaction. A crystal has several crystal
surfaces, which might have totally different mo-
lecular structures. Crystal faces are identiRed by Miller
indices, which indicate how the faces are cut from the
crystal unit cell. Figure 3 shows a unit cell and two
crystal surface structures for the compound RDX
(cyclotrimethylene trinitramine) with different
Miller indices. Since the molecular structures are very
different, the interactions of the impurity with the
surfaces might also be very different. The growth
of one face might be blocked because of strong interac-
tions, while the other face can grow freely because the
interactions are only minor. This can cause face-speci-
Rc growth retardation.
Prediction of the Molecular Structure
of the Crystalline Interface
A morphology can be geometrically constructed by
drawing planes with an orientation (hkl) having a dis-
tance from the origin proportional to the growth rate.
The central volume enclosed by the set of planes is the
growth form. A face with a large growth rate is
positioned far from the origin, and thus its surface
area will be small. In contrast, a face with a small
growth rate will be close to the origin and have a large
surface area (Figure 4). This means that the morpho-
logy is determined by the slowest growing (hkl) faces
and that a morphology prediction can be made if the
relative growth rates of all the different (hkl)
faces are known.
Figure 4 also shows the effect of a growth-
retarding impurity. Before a crystal face can grow the
adsorbed impurity has to be removed from the sur-
face. This takes energy, and thus the growth rate
decreases. When face-speciRc adsorption occurs, sur-
face concentrations of the impurity vary from face to
face. The growth retarding effect will then also
vary from face-to-face. When face-speciRc adsorption
occurs, the relative growth rates may change, as
shown in Figure 4.
Nowadays there are several general prediction
methods for vacuum morphologies. Here vacuum
means that the inSuence of the solvent or the melt is
not taken into account. Donnay and Harker’s law
states that the importance of a crystal face decreases
with interplanar distance dhkl. A morphology can be
predicted if the interplanar distance is assumed to be
inversely proportional to the relative growth rate of
the corresponding crystal face. A prediction method
that takes into account the anisotropic energies in the

Table 2 Details of additive effects on product properties

Effect on Interaction with Addition Effect on face

product property
Specific All faces Before During After Blocking Template
face crystallization crystallization crystallization growth induced
growth

Anticaking ; ; ;
Antiscaling ; ; ; ;
Morphology ; (;) ; ;
Polymorphism ; ; ;
Crystal size (;) ; ; ;
distribution
934 II / CRYSTALLIZATION / Additives: Molecular Design
Figure 2 The interaction between the surface and the impurity
is the key to the effect of the additive in the crystallization process.
This effect translates to changes in the growth and nucleation
behaviour of the crystal compound, which in turn affect the crystal
quality and process parameters and thus the product properties
change.
crystal unit cell is the attachment energy method. The
attachment energy is deRned as the energy release per
growth unit upon adding a growth slice of thickness
dhkl onto the corresponding crystal surface. The
assumption that the attachment energy is linearly
proportional to the relative growth rate of the
corresponding crystal surface gives the morphology
prediction.
The Design Method
These vacuum morphology prediction methods do
not take into account the actual growth processes
taking place on a molecular scale at the crystal surfa-
ces. Our goal is to predict the effect of an impurity
that affects these growth processes. Such a method
should include calculations of the interaction energy
between the impurity and the surface. This interac-
tion energy affects the molecular growth pro-
cesses taking place at the surface.
To make such a prediction, knowledge must Rrst be
obtained about the crystal surface structures present.
Simple energy calculations determine the mor-
phologically important surface structures. Then more
sophisticated calculation methods such as molecular
mechanics, molecular dynamics, Monte Carlo tech-
niques or even quantum mechanical calculation
methods are needed to calculate the interaction ener-
gies of the morphologically important crystal surface
structures with the impurity. These interaction ener-
gies can be translated into growth rates of the surface
structures by assuming a model which interrelates
growth rate and interaction energy. The general
method for the design of additives is shown in Fig-
ure 5.
A good way of determining surface structures and
their morphological importance is by performing
a periodic bond chain (PBC) analysis. A PBC is an
uninterrupted chain of strong bonds between growth
units in which periodicity is based on the unit cell
parameters and symmetry and which is stoichiomet-
ric with regard to the unit cell content. Two sets of
intersecting PBCs make a connected net. The connec-
ted net can be considered a (hkl) growth slice or
growth layer. If a surface does not contain a con-
nected net, the surface cannot grow with a layer
growth mechanism and is rough. This means that
the relative growth rates of (hkl) surfaces that do
not contain connected nets, are very large. These
surfaces are not present on the crystal morphology, as
a surface needs to contain a connected net to be
present. If a surface contains more than one connec-
ted net (slices at different heights) more surface
structures are possible for one crystal surface. Energy
calculations should determine which of the surface
structures actually occurs on the surface. However, it
is important to remember that the morphologically
important surface structure for a particular (hkl)
surface may change because of interactions with
impurities.
The morphologically important surface structures
can be determined by calculating the attachment en-
ergy Ea of a surface slice containing a connected net
from the bond energies in the slice (the slice energy
Esl) and the lattice energy Ecr. The attachment ener-
gies of all the possible surface structures can be used
to determine the relative growth rates of these surface
structures:
RJEa"Ecr!Esl
This results in morphologically important crystal sur-
face structures for crystals grown from systems with-
out boundary layer inSuences such as solvent interac-
tions with the surface structures.
 II / CRYSTALLIZATION / Additives: Molecular Design 935

Figure 3 The RDX (cyclotrimethylene trinitramine) unit cell (top left), and the (2 0 0) and (0 2 0) slices, indicating their surface
structure. For the (2 0 0) slice the connected net is given on the right of the molecular view of the slice while for the (0 2 0) slice the
connected net is given below the molecular view of the slice, as they occur in the unit cell. The surfaces are given perpendicular to the
paper. The dotted line in the molecular view of the surfaces indicates the surface.

Commercial software packages nowadays have
several tools to calculate interaction energies of one
or more impurity molecules on these surface struc-
tures.
Once the interaction energies have been calculated,
they should be related to the growth rates of the
crystal surface structures. The models used for this
relation are generally based on the assumption that
interaction reduces the growth rate. First the impurity
has to be removed from the surface before the face
can grow. This takes energy and thus the growth rate
decreases. An energy correction term Es for the vac-
uum attachment energy can be introduced, which is
a function of the interaction of the impurity and the
Sat crystal surface structure:

RJE
a"Ea!Es

In most cases only the interaction of a single ion,

Figure 4 The faces with the smallest growth rates determine
the morphology. A growth rate decrease (R 3PR
3) because of
a face-specific interaction of the foreign compound causes a mor-
phologically more important surface and thus a change in the
morphology.
complex or molecule with the different crystal
surfaces has to be considered. By determining the
strongest interactions the corrected attachment ener-
gies can be calculated and changes in the morphology
can be predicted.
By dynamically simulating a solvent layer on a sur-
face structure (with either a molecular dynamics or
a Monte Carlo procedure), the energy changes in-
duced in the solvent layer by the presence of the
936 II / CRYSTALLIZATION / Additives: Molecular Design
Figure 5 Method of calculating the effect of an impurity on the growth rates of the different crystal faces.
crystalline interface can be estimated, and thus the
interaction energy Es can be calculated. The proced-
ure has to be repeated for all important faces. Though
this procedure is somewhat tedious, it yields impor-
tant insights as to how a solvent can inSuence the
morphology.
Tailor-Made Additives
Tailor-made additives are usually organic additives
that are especially designed to aid in the crystalliza-
tion of a particular organic crystal compound. The
additive can be designed to work only on a very
speciRc crystal surface. Again, the additive works by
blocking the growth of this crystal surface, making
the surface morphologically more important. A char-
acteristic tailor-made additive is very similar to
a building unit. The interaction energies needed for
adsorption of the building unit and the additive to the
surface are similar. Therefore if the additive is adsor-
bed, it will be relatively hard to remove it from the
surface. To retard the growth of the crystal surface,
the additive molecule needs a functional group that
differs from the growth unit. The functional
group may be chosen so that the additive molecule
still resembles the building unit, but is smaller. The
tailor-made additive is then of the disruptive type. If,
because of its functional group, the additive molecule
is much larger than the building unit, the additive is of
the blocking type (Figure 6).
The attachment energy of a growth unit that be-
longs to a certain crystal surface can be assumed to be
linearly proportional to the growth rate of that crys-
tal face. If the slice to be attached contains an addi-
tive, the slice energy changes. However, this change
will be small if the additive strongly resembles the
building unit (as is the case in both Figure 6B and
Figure 6C: Esl+Edsl+Ebsl). Slices with tailor-made
additives will also have a similar attachment energy:
Ea+Ed1 a +Ea . The difference between crystalli-
b1
zation with and without a tailor-made additive is the
attachment energy of a subsequent normal slice onto
the slice with the additive in it. This attachment
energy (Ed2 b2
a , Ea ) is decreased because of the func-
tional group of the additive.
A disruptive type of additive will cause a decrease
in the attachment energy Ed2 a compared with the nor-
mal attachment energy Ea because some bonds be-
tween the attaching slice and the surface cannot be
made. Because the attachment energy decreases, the
growth rate of the crystal surface decreases. During
crystallization the disruptive additive can be incor-
porated in the lattice.
A blocking type of additive will cause a very large
decrease in the attachment energy Eb2 a compared with
the normal attachment energy Ea because the func-
tional group of the additive interferes with the attach-
ing slice. Either the additive has to be removed from
the surface or vacancies have to be introduced in
the attaching slice to allow the face to grow. This
takes large amounts of energy, a large decrease in
attachment energy, and thus leads to growth block-
ing. Instead of having a larger functional group,
Figure 6 Tailor-made additives can act as growth disrupters or
growth blockers.
 II / CRYSTALLIZATION / Additives: Molecular Design
a blocking additive can also have a functional group
that is differently charged from the growth unit,
and therefore rejects an incoming growth layer.
This approach can also be used for reverse engin-
eering to Rnd an impurity in the system that causes
problems such as incorporation in the lattice or inclu-
sion formation, especially when synthesis by-prod-
ucts that strongly resemble the building units are
known to be present as impurities. Disruptive impu-
rities, which have a tendency to incorporate into the
lattice, can be distinguished from blocking impurities,
which can cause macrosteps and thus inclusion
formation. Inclusion formation is the incorporation
of pockets of mother liquor into the crystal. A crystal
grown from a solution containing a blocking impu-
rity is shown in Figure 7. Very large macrosteps are
visible on the crystal, and small inclusions also appear
in the interior of the crystal.
Incorporation of Impurities
Impurities obviously reduce the purity of the crystal-
line product and are seldom beneRcial to the crystalli-
zation processes or the product characteristics. As
with the design of additives, it is important to under-
stand how the impurities are incorporated in the
crystalline product to reduce their uptake.
In general an impurity is included in the crystalline
product through either direct incorporation in the
lattice (solid solubility), the formation of inclusions,
surface adsorption or poor washing of agglomerates.
These incorporation mechanisms will be discussed in
more detail.
Solid solutions can only be formed when the impu-
rities mimic the main compound in form and charge.
Figure 7 An RDX (cyclotrimethylene trinitramine) crystal grown
from the solvent cyclohexanone. A blocking type impurity in the
system causes very large macrosteps on a particular surface and
inclusions in the interior of the crystal.
937
They are especially likely to occur with organic crys-
tals where closely related by-products are formed
during synthesis of the product. Examples are 2,6-
dimethylnaphthalene in naphthalene, cyclohexanone
in caprolactam and bromide in sodium chloride. This
phenomenon can only occur when the crystallization
enthalpy of the impurity is very close to that of the
main compound. If that is the case, not only will the
impurity be easy to include in the crystal lattice, but
the growth will not be retarded. In other words,
compounds that form high concentration solid solu-
tions have a very limited inSuence on the growth of
the main compound. If the impurity resembles the
main compound to a lesser extent the solid solubility
will decrease and growth will be hindered. The free
enthalpy involved in the incorporation of polymers or
polyelectrolytes, which are the most effective ad-
ditives in mineral scale prevention, is so large that
they are not incorporated in the lattice by direct
substitution for molecules or ions of the main com-
ponent.
A second incorporation mechanism is inclusion
formation. Inclusions can be formed in several ways.
The most common mechanism in industrial crystalli-
zation involves imperfect regrowth of corners
damaged by attrition. Pockets of mother liquor are
then trapped. The concentration of impurity in the
inclusion is then directly related to the concentration
in the mother liquor. The overgrowth of poisoned
areas is another type of inclusion formation, and
occurs at higher driving forces for crystallization.
This is essentially the only way to overcome the
blocking effects of polymers and polyelec-
trolytes. As discussed previously, the concentration of
polymers or polyelectrolytes in solution is low be-
cause these additives are effective at low concen-
trations and most of the additive is adsorbed on the
crystal surface. The amount of additive in the product
will therefore not increase much, but of course the
amount of other impurities present in the mother
liquor, such as solvent, will increase.
A third incorporation mechanism is adherence to
the surface. ‘Normal’ impurities, which cannot be
incorporated in the lattice, can be washed off
fairly easily. Tailor-made additives, however, are
designed to adhere to the surface and are difRcult
to remove. Partial dissolution of the crystals is not
an effective solution as these additives may also
hamper the dissolution process. Washing is therefore
very useful for the removal of ‘normal’ impurities,
but less so for tailor-made additives.
The formation of voids in agglomerates on impu-
rity uptake is a fourth mechanism of impurity uptake.
In conclusion, impurities that do not hinder crystal
growth can be effectively removed from the
938 II / CRYSTALLIZATION / Additives: Molecular Design
product through a combination of careful crystalliza-
tion and after treatments (washing, etc.). Compounds
such as tailor-made additives are much harder to
remove.
Case Studies
To illustrate the general principles four cases will be
discussed. These cases show the variety of possible
applications of additives and the problems sometimes
associated with the presence of impurities in the
mother liquor. Scaling inhibition is demonstrated us-
ing barium sulfate as an example, diesel fuel additives
illustrate how the crystal size distribution can be
changed, crystallization of -lactose is a good
example of self-poisoning, and the template directed
crystallization of calcium carbonate is also discussed.
Scaling problems are often encountered in oil pro-
duction. The two major scaling problems involve gas
hydrates and mineral scaling. Gas hydrates occur
when methane and water crystallize under high pres-
sure to form clathrate structures; this can happen in
gas pipes on the bottom of the sea. These crystals can
scale and block the pipes. Nowadays much research
effort is put into the design of additives to pre-
vent these clathrate crystals from blocking the pipes.
Mineral scaling occurs in secondary oil recovery.
When the immediate vicinity of a borehole becomes
depleted in oil, water is injected in the surrounding
rock strata to push the remaining oil from the pores
towards the borehole. The pores also contain water
that at these deep levels contains a high concentration
of barium. When seawater (which contains large
amounts of sulfate) is injected, barium sulfate precipi-
tates and the pores silt up. In the worst case the pores
are plugged and a new hole must be drilled, which is
a very costly operation. Scale formation is inhibited
by the addition of low amounts (ppm levels) of poly-
electrolytes, for instance polycarboxylates and poly-
sulfonates. Experiments have shown that the molecu-
lar structure of the polyelectrolyte determines the
effectiveness of the additive. Polyelectrolytes
containing sulfonate or phosphonate groups, which
closely resemble the sulfate groups in the crystal lat-
tice, are more effective than polyelectrolytes con-
taining only carboxylate groups. Also the way the
sulfonate groups are attached to the backbone is
important: molecules with the sulfonate group
directly attached are more effective than molecu-
les with the sulfonate indirectly attached. The length
of the backbone allows each sulfonate group to re-
place a sulfate in the lattice, thus binding the additive
very strongly to the crystal surface, in a zipper-like
manner, and inhibiting growth. For instance,
0.1 ppm of a polymaleic acid (polyvinyl sulfonic acid
at a pH of 3.8) decreases the growth rate of barium
sulfate by nearly two orders of magnitude (depending
on the supersaturation).
Waxes in diesel oil can crystallize at low temper-
atures. The crystals clog up the fuel Rlters in cars and
prevent the motor from starting. To prevent this
problem either wax crystallization should be pre-
vented or only very small crystals should be allowed
to form, which can pass through the fuel Rlter with-
out any problem. As the former option is not very
practical (it requires either removal of the waxes or
storage of the fuel at higher temperatures) the only
option is to add additives that block crystal growth.
The design of an additive for such a purpose has been
carried out using xylene containing n-C32H66 as
a model fuel. A copolymer of fumarate vinyl acetate
with different alkyl side chain lengths was used
as an additive. The effective concentration at
which the additive started to inSuence the growth of
n-C32H66 was determined at a Rxed supersaturation as
a function of the alkyl side chain length. A clear
relationship between the side chain length and the
growth inhibition was shown, with a maximum ef-
fectiveness when the chain length was very close to
the chain length of the solute. This shows that the
structural match between the additive and the solute
is critical for the effectiveness of the additive.
-Lactose, a milk sugar, is a disaccharide contain-
ing a galactose and a glucose ring. Under inSuence of
an acid, the -glucose ring can be opened and closed
again either in the  or the  form. In the latter form it
is a tailor-made additive. The galactose moiety re-
sembles lactose, but the different position of the
hydroxyl group ( instead of ) hinders subsequent
growth. The ‘-side’ of the crystal can therefore be
poisoned so the -lactose molecules cannot attach to
the crystal, and only one side is blocked. This results
in the typical tomahawk shape of the crystals (Fig-
ure 8).
Figure 8 The -lactose hydrate crystal grown in the presence
of -lactose. The (0 1 0) surface on the right is blocked because
-lactose can adsorb on the surface, causing a blocking impurity
as shown in Figure 6C, while on the left the (0 1 0) surface is not
blocked because the -lactose can not adsorb on the surface.
 II / CRYSTALLIZATION / Additives: Molecular Design
A substance that is capable of crystallizing into
structurally different but chemically identical
crystalline forms exhibits polymorphism. A very com-
mon compound that can form polymorphs is water.
Under atmospheric pressure and slightly below 03C
ice is formed with a density lower than that of water,
so the ice Soats on the water. Under higher pressure
the ice can crystallize into polymorphs with higher
densities that sink. Different polymorphs crystal-
lize under different conditions. However, these
conditions may be almost equal for rather complex
organic compounds such as pharmaceuticals. In the
pharmaceutical industry a strong desire exists to con-
trol the crystallization of polymorphs for a number of
reasons. The polymorphic form of a pharmaceutical
inSuences its effect and its lethal dosage. The
crystallization processes of particular polymorphs are
therefore protected by patents.
One way of controlling the crystallization of the
correct polymorph is by means of template-directed
nucleation of the polymorph. The template mimics
a crystal face of the desired polymorph and nuclei of
that polymorph can form onto the template. Up to
now attention has mainly focused on templates con-
sisting of Langmuir}Blodgett (LB) layers.
Calcite and vaterite are polymorphs of the com-
pound calcium carbonate (calcite is the polymorph
that forms under ambient conditions). Stearic acid
spreads at the air}water interface, which when com-
pressed as a LB monolayer is highly structured. When
the acid group is ionized at higher pH the monolayer
surface facing the solution consists of carboxylates
that can bind calcium ions from the solution. At
higher calcium concentrations calcite grows from the
(1 1 0) face under the monolayer, while at lower
calcium concentrations vaterite grows from the
(0 0 1) face under the monolayer. It is suggested that
the surface concentration of calcium under the
monolayer is important: at high concentrations the
calcium ions bonded to the monolayer mimic the
(1 1 0) face of calcite while at lower concentrations
they mimic the (0 0 1) face of vaterite, which is
less densely packed with calcium ions. Many
examples of template-induced growth are found in
nature, as mentioned before and only recently has
this technique for controlling polymorphs been
exploited.
Concluding Remarks
Additives can be used to improve the handling and
quality of crystal products from an industrial crystal-
lization process. The large number of effects that
an additive can have on the handling and the quality
939
all come down to the interaction between the additive
and the different crystal surfaces. A general
method was given for the design of additives, which
involves gaining knowledge of the morphologically
important crystal surfaces and on the interaction be-
tween these surfaces and the additive.
Further Reading
Allen MD and Tildesley DJ (1987) Computer Simulation of
Liquids. Oxford: Clarendon Press.
Beiny DHM, Mullin JW and Lewtas K (1990) Crystalliza-
tion of n-dotriacontane from hydrocarbon solution with
polymeric additives Journal of Crystal Growth 102:
801d806.
Clydesdale G, Roberts KJ and Docherty R (1994) Model-
ling the morphology of molecular crystals in the pres-
ence of disruptive tailor-made additives. Journal of
Crystal Growth 135: 331}340.
Clydesdale G, Roberts KJ, Lewtas K and Docherty
R (1994) Modelling the morphology of molecular crys-
tals in the presence of blocking tailor-made additives.
Journal of Crystal Growth 141: 443}450.
de Goede R (1992) Improvement of melt crystallizations
efRciency for industrial applications. In: Pilavachi
PA (ed.) Energy EfTciency in Process Technology.
Elsevier Applied Science.
Donnay JHD and Harker D (1937) American Mineralogist
22: 446.
Frenkel D and Smit B (1996) Understanding Molecular
Simulation. New York: Academic Press.
Hartmann P and Bennema P (1980) The attachment energy
as a habit controlling factor. I. Theoretical consider-
ations. Journal of Crystal Growth 49: 145}156.
Hurle DTJ (ed.) (1993) Handbook of Crystal Growth, vols
1A}3B. Amsterdam: Elsevier Science.
Jacquemain D, Wolf SG, Leveiller F, Deutsch M, Kjaer K,
Als-Nielsen J, Lahav M and Leiserowitz L (1992)
Angewandte Chemie International Edition English 31:
130}152.
Lin CH, Gabas N, Canselier JP and Pèpe G (1998) Predic-
tion of the growth morphology of aminoacid crystals in
solution. I. -Glycine. Journal of Crystal Growth 191:
791}802.
Mann S and Heywood BR (1988) Antifreeze polypeptides
come out of the cold. Nature 335 (6190): 502}503.
Mann S, Heywood BR, Rajam S and Birchall JD (1988)
Nature 334: 692d695.
Mann S, Heywood BR, Rajam S and Birchall JD (1989)
Interfacial control of nucleation of calcium carbonate
under organized stearic acid monolayers. Proceeding of
the Royal Society of London, A 423: 457.
Mann S, Heywood BR, Rajam S, et al., (1990) Advanced
Materials 2: 257.
Mersmann A (ed.) (1995) Crystallization Technology
Handbook. New York: Marcel Dekker, Inc.
940 II / CRYSTALLIZATION / Biomineralization
Mullin JW (1993) Crystallization, 3rd edn. Oxford: Butter-
worth-Heineman Ltd.
Myerson AS, ed. (1999) Molecular Modeling Applications
in Crystallization. Cambridge: Cambridge University
Press.
Biomineralization
D. Volkmer, University of Bielefeld, Bielefeld, Germany
Copyright ^ 2000 Academic Press
Many organisms have developed sophisticated
strategies to direct the growth of the inorganic con-
stituents of their mineralized tissues. Active control
mechanisms are effective at almost all levels of struc-
tural hierarchy, ranging from the nanoscopic regime
} the nucleation of a crystallite at a speciRc site } up to
the macroscopic regime, where the biophysical prop-
erties of the mineralized tissue have to be matched to
a certain function.
Among the many open questions, one of the most
challenging scientiRc problems is to gain insights into
the molecular interactions that occur at the interface
between the inorganic mineral and the macromolecu-
lar organic matrix. Biogenic crystals often express
exceptional habits that are seemingly unrelated to the
morphology of the same type of crystals when grown
under equilibrium conditions. For the most wide-
spread calciRed tissues it is frequently assumed that
a structurally rigid composite matrix consisting of
Rbrous proteins and thereon adsorbed acidic macro-
molecules acts as a supramolecular blueprint that
templates nucleation of the inorganic phase. Sub-
sequent crystal growth proceeds within a specialized
compartment which encloses a suitable aqueous
microenvironment. The particular composition of
solutes, which often comprises a complex mixture of
dissolved electrolytes and macromolecules, has
a strong inSuence on the morphology of the crystals.
In the course of mineral deposition, growth modiRers
may interact with the maturing crystal in different
ways: dissolved macromolecules may be adsorbed
onto speciRc crystal faces, thus slowing down or
inhibiting deposition rates along certain crystallo-
graphic directions. Adsorbed macromolecules may be
completely overgrown by the mineral to produce lat-
tice defects or to introduce discontinuities in the crys-
tal texture.
Efforts in trying to separate and mimic aspects of
these complex interactions with simple model sys-
tems will help to improve our understanding of crys-
tallization processes that are under biological control.
Sherwood J (1969) Defects in organic crystals. Molecular
Crystals and Liquids 9: 37.
Sloan ED Jr (1990) Clathrate Hydrates of Natural Gases.
New York: Marcel Dekker Inc.
A proRtable knowledge transfer in the direction of
biologically inspired design strategies for building
new and improved composite materials can be pre-
dicted for the near future.
The following account of biomineralization fo-
cuses on two special topics of this wide research
Reld, namely ferritin and mollusc shell mineraliz-
ation, which are considered here as illustrative
examples. For a more comprehensive survey which
includes further important types of biominerals
(e.g. bone or biogenic structures made of amorphous
silica) the reader should consider one of the many
excellent monographs and review articles on the sub-
ject.
Ferritin: From Iron Storage
to Nanoparticle Synthesis
Mineral deposition in the iron storage protein (fer-
ritin) may be regarded as an archetypal biological
model for the formation of a nanocrystalline mineral
phase within a conRned space. The structure and
function of ferritin have been reviewed in great detail.
Ferritin consists of an oligomeric protein shell
(apoferritin) and a core of poorly crystalline Fe(III)
oxyhydroxide (presumably ferrihydrite, 5 Fe2O3 )
9 H2O). Iron is temporarily stored within and re-
leased from the central cavity of the encapsulating
protein shell. The availability of several high resolu-
tion three-dimensional structures of apoferritins ori-
ginating from different organisms provides a reliable
basis to discuss possible pathways of iron biomineral-
ization. Current biomimetic strategies to achieve
similar properties include mineralization in oil}
water microemulsions, block copolymer micelles,
or biotechnologically produced capsule-forming
proteins.
Apoferritin Structure and Biological
Function
Apoferritins with different amino acid compositions
have been isolated from eukaryots as well as